fixing bug that we need to check elf header info to match libs.

Signed-off-by: vsoch <vsoch@users.noreply.github.com>

README.md

@@ -82,6 +82,21 @@ __cxa_atexit    __cxa_finalize
 
 ```
 
+The defaults above show the console. DIfferent formats are shown below (under development).
+
+#### Text
+
+```bash
+$ elfcall gen data/libfoo.so --fmt text
+/usr/lib/x86_64-linux-gnu/libstdc++.so.6           LINKSWITH            libm.so.6
+/usr/lib/x86_64-linux-gnu/libstdc++.so.6           LINKSWITH            libc.so.6
+/usr/lib/x86_64-linux-gnu/libstdc++.so.6           LINKSWITH            ld-linux-x86-64.so.2
+/usr/lib/x86_64-linux-gnu/libstdc++.so.6           LINKSWITH            libgcc_s.so.1
+/usr/lib/x86_64-linux-gnu/libstdc++.so.6           EXPORTS              _ZNSt8ios_base4InitC1Ev
+/usr/lib/x86_64-linux-gnu/libstdc++.so.6           EXPORTS              _ZNSt8ios_base4InitD1Ev
+```
+
+
 Note that this is under development, and eventually we will have different graph generation
 options (right now we print to the screen).
 
@@ -92,19 +107,13 @@ You can also generate a tree of the library paths parsed:
 ```bash
 $ elfcall tree data/libfoo.so
 libstdc++.so.6                 [x86_64-linux-gnu.conf]
-   libm.so.6                   [i386-linux-gnu.conf]
-      libc.so.6                [i386-linux-gnu.conf]
-         ld-linux.so.2         [i386-linux-gnu.conf]
-      ld-linux-x86-64.so.2     [x86_64-linux-gnu.conf]
-      libgcc_s.so.1            [i386-linux-gnu.conf]
+   ld-linux-x86-64.so.2        [x86_64-linux-gnu.conf]
 ```
 
 or:
 
 ```bash
 $ elfcall tree /usr/bin/vim
-libm.so.6                      [i386-linux-gnu.conf]
-   ld-linux.so.2               [i386-linux-gnu.conf]
 libtinfo.so.6                  [x86_64-linux-gnu.conf]
 libselinux.so.1                [x86_64-linux-gnu.conf]
    libpcre2-8.so.0             [x86_64-linux-gnu.conf]
@@ -117,13 +126,8 @@ libcanberra.so.0               [x86_64-linux-gnu.conf]
    libltdl.so.7                [x86_64-linux-gnu.conf]
 libacl.so.1                    [x86_64-linux-gnu.conf]
 libgpm.so.2                    [x86_64-linux-gnu.conf]
-libdl.so.2                     [i386-linux-gnu.conf]
 libpython3.8.so.1.0            [x86_64-linux-gnu.conf]
    libexpat.so.1               [x86_64-linux-gnu.conf]
-   libz.so.1                   [x86_64-linux-gnu.conf]
-   libutil.so.1                [i386-linux-gnu.conf]
-libpthread.so.0                [i386-linux-gnu.conf]
-libc.so.6                      [i386-linux-gnu.conf]
 ```
 
 

elfcall/client/__init__.py

@@ -61,6 +61,14 @@ def get_parser():
         description="generate the symbol callgraph.",
         formatter_class=argparse.RawTextHelpFormatter,
     )
+    gen.add_argument(
+        "--fmt",
+        "-f",
+        help="graph format to generate",
+        choices=["text", "ge", "gexf", "console"],
+        default="console",
+    )
+
     tree = subparsers.add_parser(
         "tree",
         description="generate a library tree",

elfcall/client/gen.py

@@ -12,4 +12,4 @@ def main(args, parser, extra, subparser):
     if not args.binary:
         logger.exit("You must provide one or more binaries to parse.")
     cli = BinaryInterface(args.binary[0], quiet=args.quiet)
-    cli.gen()
+    cli.gen(fmt=args.fmt)

elfcall/main/client.py

@@ -66,15 +66,14 @@ def tree(self, binary=None):
         self.reset()
         self.ld.parse()
 
-        # TODO will we ever want to match:
-        # e_ident[EI_DATA], e_ident[EI_CLASS], e_ident[EI_OSABI], e_ident[EI_ABIVERSION],
-        # e_machine, e_type, e_flags and e_version.
-        # https://github.com/vsoch/elfcall/issues/1
-
-        results = self.recursive_find(binary)
+        # Load original binary - we need to match elf attributes here
+        original = elf.ElfFile(binary)
+        results = self.recursive_find(binary, original=original)
         self.library_tree(results)
 
-    def recursive_find(self, lib, root=None, needed_search=None, seen=None, level=0):
+    def recursive_find(
+        self, lib, original, root=None, needed_search=None, seen=None, level=0
+    ):
         """
         recursively find needed paths, keep track of hierarchy
         """
@@ -113,6 +112,10 @@ def recursive_find(self, lib, root=None, needed_search=None, seen=None, level=0)
 
                 libelf, src, already_seen = self.find_library(path, search_paths)
 
+                # If it does not match the arch and elf type, ignore
+                if not original.matches(libelf):
+                    continue
+
                 # We might get back a soname instead we've already seen
                 if already_seen:
                     continue
@@ -149,6 +152,7 @@ def recursive_find(self, lib, root=None, needed_search=None, seen=None, level=0)
                 root=next["root"],
                 needed_search=next["needed"],
                 level=next["level"],
+                original=original,
             )
 
         return root
@@ -257,6 +261,10 @@ def parse_binary(self, binary):
                 # This will return loaded ELF, if found, otherwise None
                 libelf, _, already_seen = self.find_library(path, search_paths)
 
+                # If it does not match the arch and elf type, ignore
+                if not e.matches(libelf):
+                    continue
+
                 # We might get back a soname instead we've already seen
                 if already_seen:
                     continue
@@ -277,7 +285,13 @@ def parse_binary(self, binary):
                 for name, symbol in imported.items():
                     if name in exported_contenders:
                         logger.debug("Found %s -> %s" % (name, path))
-                        found[name] = {"lib": lib, "linked_libs": libelf.needed}
+                        found[name] = {
+                            "lib": {
+                                "realpath": libelf.realpath,
+                                "fullpath": libelf.fullpath,
+                            },
+                            "linked_libs": libelf.needed,
+                        }
                         found[name].update(symbol)
                         to_removes.append(name)
                 for to_remove in to_removes:

elfcall/main/elf.py

@@ -72,6 +72,22 @@ def needed(self):
                     self._needed.append(tag.needed)
         return self._needed
 
+    def matches(self, libelf):
+        """
+        Compare an elf to another lib's elf type, arch, and header metadata.
+        Return True if they match, False otherwise.
+        """
+        # Note that including EI_OSABI
+        # e_ident[EI_DATA], e_ident[EI_CLASS], e_ident[EI_OSABI], e_ident[EI_ABIVERSION],
+        for field in ["EI_CLASS", "EI_DATA", "EI_ABIVERSION"]:
+            if self.header["e_ident"][field] != libelf.header["e_ident"][field]:
+                return False
+        # e_machine, e_type, e_flags and e_version.
+        for field in ["e_type", "e_machine", "e_flags", "e_version"]:
+            if self.header[field] != libelf.header[field]:
+                return False
+        return True
+
     @property
     def operating_system(self):
         return self.header["e_ident"]["EI_OSABI"]

elfcall/main/graph/text.py

@@ -22,16 +22,16 @@ def generate(self):
 
 class Text(GraphBase):
     def generate(self):
-        logger.info("Output will be written to %s" % self.outfile)
         if self.outfile != sys.stdout:
+            logger.info("Output will be written to %s" % self.outfile)
             fd = open(self.outfile, "w")
         else:
             fd = self.outfile
 
         # Record linked dependencies
         for filename, symbols in self.organized.items():
             for linked_lib in self.linked_libs[filename]:
-                fd.write("%s LINKSWITH %s\n" % (filename, linked_lib))
+                fd.write("{:50} {:20} {}\n".format(filename, "LINKSWITH", linked_lib))
 
         # We only care about each library imports (exported from another)
         exported = self.get_exported()
@@ -40,14 +40,13 @@ def generate(self):
         for symbol in exported:
             placeholder = self.generate_placeholder()
             self.symbol_uids[symbol[0]] = placeholder
-            fd.write("%s EXPORTS %s\n" % (filename, symbol[0]))
 
         # store which files use which symbols
         for filename, metas in self.organized.items():
             for meta in metas:
                 symbol = meta["name"]
                 placeholder = self.symbol_uids[symbol]
-                fd.write("%s USES %s\n" % (filename, symbol))
+                fd.write("{:50} {:20} {}\n".format(filename, "EXPORTS", symbol))
 
         if self.outfile != sys.stdout:
             fd.close()